The present invention relates to a probe device for performing tests on an object to be tested, such as an IC chip that forms a semiconductor device.
During the process of fabricating a semiconductor device, once the processing of the wafer itself has ended and the IC chips within the wafer have been completed, electrical measurements called probe tests are performed to determine the quality of the IC chips while they are still in semiconductor wafer form, in order to investigate short- and open-circuits between electrodes in the electrode pattern and the input-output characteristics of each IC chip. The wafer is then divided into individual IC chips and the chips that passed these tests are packaged and subjected to further probe tests before their quality as completed products is determined.
A prior art probe device of this type is provided with a wafer mounting stand that is capable of moving in one direction within a horizontal plane (X direction) and another direction that is either perpendicular thereto (Y direction) in that plane or is in the vertical direction (Z direction), and of rotation about a vertical axis. A probe card, which is provided with probe needles that are arrayed to correspond with electrode pads on the chips in the wafer, is arranged above the wafer mounting stand. The configuration is such that, after the wafer mounting stand has been moved to align the electrode pads of the IC chips in the wafer with the probe needles, the probe needles are bought into contact with the electrode pads, the probe needles are placed in electrical contact with a test head via a contact ring comprising pogo pins and so on, and then the quality of each of the IC chips is determined by performing electrical tests using frequencies corresponding to the operating speed of the ICs, for example.
To ensure accurate electrical measurements, the probe needles must be made to come into reliable contact with the electrode pads, and therefore it is necessary to accurately position the electrode pads of the IC chips on the wafer with respect to the probe needles beforehand. Therefore, a device such an optical unit for detecting wafer patterns is provided in a position separated from the test head and the wafer is positioned under the optical unit by moving it in the X and Y directions. The position of the wafer mounting stand is adjusted while the electrode pads are observed through the optical unit, and thus the wafer is positioned in the X, Y, and .theta. directions, for example.
A burn-in test that detects IC chips that will fail under severe conditions is usually performed after the chips are packaged, but recently it has become common to perform this test while the chips are still in wafer form. In this case, a heater is built into the wafer mounting stand and the measurements are performed while the temperature of the chips is adjusted thereby within a range of, for example, -40.degree. C. to +150.degree. C.
However, as chip sizes become smaller, circuitry becomes more densely integrated, and processing speeds become faster, and as the semiconductor wafers on which large numbers of chips are formed are tending to become bigger in diameter, various defects with the probe device are becoming obvious. For example, the increase in diameter of the wafers makes it necessary to provide a large space for moving the wafer mounting stand, including an area for positioning the wafer, which means that the measurements at low and high temperatures will be affected to some extent by the ambient temperature, and also particles are likely to be generated by the movement of the wafer mounting stand. Another problem concerns LSIs for supercomputers that have operating speeds that are extremely high, such as several tens to several hundreds of MHz. If operating speeds are increased in this manner, the frequency of the signal pulses used during the measurements is also increased, making the device susceptible to electrical noise from the outside. The effects of electrical noise can be suppressed by using a coaxial cable for the pogo pins of the contact ring, but an electrical shielding structure cannot be used for the ends of the pogo pins or the probe needles. These problems have raised the further problem of instability in the electrical measurement of chips.
As chip sizes become smaller and circuitry becomes more densely integrated, the numbers of electrode pads increase and also the sizes of these pads become smaller and the spacing between them becomes tighter, and thus there is the problem that the pitch of the probe needles is approaching a physical limit.
There has recently been studies into the use of probe cards on which are formed bumps that are conductive protrusions of a material such as 18K gold or copper on a thin-film substrate formed of a resin such as a polyimide, as contactors. With such a probe card, tiny bumps can be accurately formed in a predetermined array by using a technique such as printing on an insulating substrate, and there are expectation that this technique will be applied to devices in the future.
However, when a wafer is heated to about 150.degree. C. during burn-in testing, the probe card is also heated and thus the probe card is likely to expand, changing the positions (coordinate positions) of the bumps on the probe card, and thus necessitating detection of these changes and repositioning of the probe card and the wafer after other tests such as normal-temperature tests.
If conventional means were to be adapted for such repositioning, a method could be considered in which two marks attached to the probe card are monitored from below by a camera, a status such as the thermal expansion status of the probe card (or rather the positional status of the bumps) is derived therefrom, and the position of the wafer is adjusted to match the array of bumps. However, the coordinate positions of the bumps extending over the region in which the bumps are arrayed cannot be determined by this method (for example, if the degree of thermal distortion varies locally), and it could happen that not all of the bumps will come into contact accurately with the electrode pads, particularly when there is a large number of bumps and their array density is high. There is also a problem in that labor-intensive work is necessary for previously measuring the coordinate positions of the bumps on the basis of monitor effects for each different type of probe card.
Another method that could be considered involves monitoring all of the bumps with a camera, but the work of checking each bump array is extremely troublesome, and it is difficult to look at all of the bumps to determine their coordinate positional status. Performing high-temperature testing after normal-temperature testing with this method is very inefficient because the same operation must be repeated in a heated status after the wafer has been positioned under normal temperature conditions.